From fd42607cd55efd4a2c634a8eaeec509a45e34297 Mon Sep 17 00:00:00 2001
From: Till Hartmann <till.hartmann@bih-charite.de>
Date: Thu, 23 May 2024 10:11:59 +0200
Subject: [PATCH] add pydantic model for cbioportal_export

---
 .../workflows/cbioportal_export/model.py      | 153 ++++++++++++++++++
 1 file changed, 153 insertions(+)
 create mode 100644 snappy_pipeline/workflows/cbioportal_export/model.py

diff --git a/snappy_pipeline/workflows/cbioportal_export/model.py b/snappy_pipeline/workflows/cbioportal_export/model.py
new file mode 100644
index 000000000..83dec48e3
--- /dev/null
+++ b/snappy_pipeline/workflows/cbioportal_export/model.py
@@ -0,0 +1,153 @@
+from __future__ import annotations
+
+from enum import Enum
+from typing import Any, Annotated
+
+from pydantic import BaseModel, ConfigDict, Field
+
+
+class MappingTool(Enum):
+    BWA = "bwa"
+
+
+class ExpressionTool(Enum):
+    STAR = "star"
+
+
+class SomaticVariantCallingTool(Enum):
+    MUTECT2 = "mutect2"
+
+
+class SomaticVariantAnnotationTool(Enum):
+    VEP = "vep"
+
+
+class FilterSet(Enum):
+    NO_FILTER = "no_filter"
+    DKFZ_ONLY = "dkfz_only"
+    DKFZ_AND_EBFILTER = "dkfz_and_ebfilter"
+    DKFZ_AND_EBFILTER_AND_OXOG = "dkfz_and_ebfilter_and_oxog"
+
+
+class Filter(BaseModel):
+    pass
+
+
+class DkfzFilter(Filter):
+    pass
+
+
+class EbFilter(Filter):
+    ebfilter_threshold: float = 2.4
+    shuffle_seed: int = 1
+    panel_of_normals_size: int = 25
+    min_mapq: int = 20
+    min_baseq: int = 15
+
+
+class Bcftools(Filter):
+    include: str = ""
+    exclude: str = ""
+
+
+class Regions(Filter):
+    path_bed: str = ""
+
+
+class Protected(Filter):
+    path_bed: str = ""
+
+
+class CopyNumberTool(Enum):
+    CNVKIT = "cnvkit"
+
+    CONTROL_FREEC = "Control_FREEC"
+    """unsupported"""
+
+    COPYWRITER = "CopywriteR"
+    """unmaintained"""
+
+
+class NcbiBuild(Enum):
+    GRCh37 = "GRCh37"
+    GRCh38 = "GRCh38"
+
+
+class Vcf2Maf(BaseModel):
+    Center: str
+    ncbi_build: NcbiBuild
+
+
+class Study(BaseModel):
+    type_of_cancer: str
+    """
+    see http://oncotree.mskcc.org/#/home
+    see also `curl https://oncotree.mskcc.org:443/api/tumorTypes | jq ".[].code"`
+    """
+    cancer_study_id: str
+    """Usually: <type of cancer id>_<pi>_<year>"""
+    study_description: str
+    study_name: str
+    study_name_short: str
+
+
+class CbioportalExport(BaseModel):
+    model_config = ConfigDict(
+        extra="forbid",
+    )
+
+    path_ngs_mapping: str | None = None
+    """When missing, no expression data is uploaded to cBioPortal"""
+
+    mapping_tool: MappingTool = MappingTool.BWA
+
+    expression_tool: ExpressionTool = ExpressionTool.STAR
+
+    path_somatic_variant: str
+    """REQUIRED (before or after filtration)"""
+
+    somatic_variant_calling_tool: SomaticVariantCallingTool = SomaticVariantCallingTool.MUTECT2
+    """mutect/scalpel combo unsupported"""
+
+    somatic_variant_annotation_tool: SomaticVariantAnnotationTool = SomaticVariantAnnotationTool.VEP
+
+    filter_set: Annotated[FilterSet | None, Field(None, deprecated="use `filter_list` instead")]
+    """
+    DEPRECATED: use `filter_list instead`.
+    Set it to an empty value when using annotated variants without filtration.
+    """
+
+    exon_list: Annotated[
+        str | None,
+        Field(
+            "genome_wide",
+            deprecated="Works together with filter_set, " "ignored when `filter_list` is selected",
+        ),
+    ]
+    """
+    DEPRECATED.
+    Works together with filter_set, ignored when "filter_list" is selected
+    """
+
+    filter_list: list[Filter] = []
+
+    exclude_variant_with_flag: str | None = None
+    """Required for Copy Number Alterations"""
+
+    path_copy_number: str | None = None
+    """When missing, no CNV data uploaded to portal. Access WES & WGS steps"""
+
+    copy_number_tool: CopyNumberTool = CopyNumberTool.CNVKIT
+
+    path_gene_id_mappings: str
+    """Mapping from pipeline gene ids to cBioPortal ids (HGNC symbols from GeneNexus)"""
+
+    vcf2maf: Vcf2Maf
+
+    study: Study
+
+    patient_info: None
+    """unimplemented"""
+
+    sample_info: Any
+    """Each additional sample column must have a name and a (possibly empty) config attached."""